Deteksi dan Klasifikasi Larva Nyamuk Menggunakan Arsitektur YOLOv8 dan MobileNetV3

Abstract

Real-time detection and classification of mosquito larvae on mobile devices still face numerous challenges in terms of accuracy and efficiency. There are some limitations in manual identification, thus it is necessary to develop a deep learning-based system to improve accuracy and efficiency speed in diagnosis. This study proposed a mosquito larva detection and classification model using YOLOv8 and MobileNetV3 on mobile devices. The objectives of this research are to improve accuracy and efficiency in identifying mosquito larvae of the Aedes and Culex genus, as well as the unknown class, which includes the Anopheles and Toxorhynchites genera, in order to support environmental health monitoring. YOLOv8 method was employed for object detection and MobileNetV3 for mosquito larva classification. The dataset used consists of images of Aedes, Culex, Anopheles, and Toxorhynchites larvae. The model was then trained and evaluated using deep learning techniques, then applied to a mobile application to automatically detect and classify larvae. The results indicate that the developed system is capable of detecting and classifying mosquito larvae with high accuracy, where YOLOv8 achieves an mAP50 of 0.986 and mAP50-95 of 0.777. At the same time, MobileNetV3 produced a classification accuracy of 0.962. In terms of efficiency, the model was able to perform real-time inference on the mobile devices with optimized processing time. Stable performance can be seen on new data, proving its potential in environmental health monitoring and supporting more effective disease vector control as well as aiding further research in the field of entomology.